Crop spatial characterization in the 18th aquifer, Spain

摘要

The results of a remotely sensed imagery classification depend on a lot of elemetns (information classes, remote sensed images, conditions of image acquisition, the spectral and spatial resolution etc.) but, mainly, how the classification algorithm is trained. In this work, a new pattern characterization is proposed. Its main property is that the spatial variability component is considered in the own pattern; that is to say, the problems inherited from the confidence boundary location of hte spectrally homogeneous zones are avoided. The corn crop spatial pattern in the 18th aquifer of hte Mancha Oriental (Spain) has been used as the test bed.